plot_confusion_matrix() 缺少 1 个必需的位置参数："y_true"

Question

我想绘制我训练模型的混淆矩阵，但当我试图绘制我的混淆矩阵时卡住了。没有。当预测值和实际值的样本长度相同时，两个参数的长度相同，但仍然抛出误差

TypeError: plot_confusion_matrix()缺少1个必需的位置参数:'y_true'

这里是我的代码供参考

# downloading VGG16 model 
img_height, img_width = 224,224
conv_base = vgg16.VGG16(weights='imagenet', include_top=False, pooling='max', input_shape = (img_width, img_height, 3))
# creating a model without freezing the layers of Model
model = models.Sequential()
model.add(conv_base)
model.add(layers.Dense(nb_categories, activation='softmax'))
model.summary()
learning_rate = 5e-5
epochs = 3
checkpoint = ModelCheckpoint("sign_classifier.h5", monitor = 'val_acc', verbose=1, save_best_only=True, save_weights_only=False, mode='auto', period=1)
model.compile(loss="categorical_crossentropy", optimizer=tensorflow.optimizers.Adam(lr=learning_rate, clipnorm = 1.), metrics = ['acc'])    
history = model.fit_generator(train_generator, 
epochs=epochs, 
shuffle=True, 
validation_data=val_generator,
callbacks=[checkpoint]
)
Y_pred = model.predict_generator(test_generator)
y_pred = np.argmax(Y_pred, axis=1)
print("test data samples",len(test_generator.classes))
print("predicted data samples",len(y_pred))
print("Shape of Predicted Value",y_pred.shape)
print("Shape of actual Value",test_generator.classes.shape)
y_true = test_generator.classes
cm = confusion_matrix(test_generator.classes, y_pred)
print("Confusion Matrix",cm)
plot_confusion_matrix(y_true, y_pred, labels = category_names, normalize=False)

以上代码的输出为

test data samples 47
predicted data samples 47
Shape of Predicted Value (47,)
Shape of actual Value (47,)
Confusion Matrix [[22  0  1]
[ 0  9  0]
[13  0  2]]

Answer 1

sklearn.metrics。plot_confe_matrix需要3个位置参数classifier,X,y_true。所以你需要传递你的模型作为分类器参数。

但是，如果您将模型作为参数传递给函数，则会出现另一个问题，您的模型将不会被识别为分类器。为了解决这个问题，我创建了一个笔记本，可能会帮助你。这里是链接

Answer 2

使用以下格式:

plot_confusion_matrix(X =test_generator.classes, y_true = y_pred, labels = category_names, normalize=False)